48 results
Filters
Settings
Now showing1 - 10 of 48
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Optical simulation framework for spatial optimization and channel characterization in intra-satellite LiFi systemsLight Fidelity (LiFi) for intra-satellite communication offers significant advantages, including flexible Assembly, Integration and Testing Phase (AIT) and reduced harness complexity. This paper presents a simulation framework for light propagation, primarily aimed at optimizing positioning, spatial structures, and system parameters for intra-satellite LiFi architectures. Initial simulations included generic scenarios such as obstacle simulation and transmitter–receiver alignment, followed by integration of a complete 3D CubeSat model. Results demonstrate the framework’s capability to support the optimization and visualization of optical communication structures, enabling virtual analysis that complements physical test setups in early development stages. The framework has been extended to incorporate diffuse reflection and simple wavelength-independent refraction, while supporting iterative reuse of simulation results and channel characteristics enabling iterative transmitter and receiver placement optimization.Publicationtype: Journal Article6 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , (DEMO) Real-time bird detection with a modular acoustic sensing platformPassive Acoustic Monitoring (PAM) enables continuous, non-invasive observation of wildlife and is widely used for biodiversity assessment. Long-term deployments require sensing platforms that are reliable, energy-efficient, affordable, and flexible for future extensions such as embedded inference.This demo paper presents a real-time bird detection system built around a modular embedded acoustic sensing platform for low-power environmental audio recording. The live demonstration shows the workflow from acoustic acquisition to automated BirdNET-based species inference. Bird vocalizations are recorded by the sensing platform, transferred to a connected host system, and analyzed using BirdNET to visualize detected species, confidence values, and temporal activity patterns in real time.The demonstration showcases an end-to-end low-power bioacoustic monitoring workflow, combining autonomous environmental audio acquisition, embedded sensing hardware, and real-time BirdNET-based species inference within a compact modular platform.Publicationtype: Conference Paper9 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , CubeSat LAB - From Hamburg to SpaceThe space sector is expanding rapidly and requires a new generation of students to recognise space as a distinct engineering discipline, rather than perceiving it as ‘rocket science’. Hamburg is known for its aviation and maritime industries, which is reflected in the education at the Hamburg University of Technology. Enabling interdisciplinary space education and especially fostering self-learning in electronics and computer science topics at the TU Hamburg is the goal of the EduSat Project of the Smart Sensors Group. This is achieved by a combination of lectures, Project-based learning and the CubeSat LAB, the latter providing students with the opportunity to build their own payload, work on a provided cubesat and expose students to the space sector. The CubeSat LAB is unique in its focus on the electronics of a cubesat, caused by a lack of pre-existing space engineering curricula at our university. This lack of theoretical space engineering knowledge is one of the challenges the LAB, and its students, faces. Further this means students have to transfer and apply their theoretical knowledge to the space domain. The goal of the LAB is to offer physical and intellectual space for students from various backgrounds and study levels to work together interdisciplinarily and get in contact with the space sector and its engineering challenges, beyond their curriculum. Furthermore, working in different subteams, communicating with each other, and solving challenges in a bigger team prepares the students for their future career. While the CubeSat LAB is part of the Smart Sensors Group, it is governed by students, whereas tutors are supervising the team and providing the interface to the group. The current student mission aims to autonomously capture the northern lights, with the help of a variety of sensors. The team is close to the end of the concept phase, working to expand participation and has founded an association to support their efforts. Further the students are participating in the REXUS program, with a LiFi-based experiment. This paper presents the structure, challenges, and first outcomes of the CubeSat LAB, presenting its role in space education at the TU Hamburg and promoting interdisciplinary engagement with the space sector.Publicationtype: Conference Paper8 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , POSTER: Discrete wavelets as zero-parameter front-end for near-sensor cardiac-interval estimation from seismocardiographyWe present the Discrete Wavelet Transform (DWT) as a zero-parameter front-end for near-sensor cardiac-interval estimation from Seismocardiography (SCG), preserving timing-critical structure under embedded constraints. Across 10 wavelet families, we benchmark reconstruction, peak-timing fidelity, and energy on an nRF52840 SoC, showing that energy scales with filter length, while signal fidelity depends on wavelet family. This yields an energy-fidelity Pareto frontier for principled wavelet selection, with ongoing work extending this representation into neural network pipelines for cardiac interval estimation with morphology-dependent inductive bias.Publicationtype: Conference Paper4 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , From sound to species: a modular acoustic sensor for biodiversity monitoringPassive Acoustic Monitoring (PAM) is widely used for biodiversity assessment and environmental monitoring of wildlife and soundscapes. Long-term field deployments require sensing platforms that are compact, low-power, robust, and cost-efficient while remaining flexible for future extensions.This work presents a modular embedded acoustic sensing platform for autonomous environmental audio recording. The system separates power management, processing and data logging, and audio acquisition into dedicated subsystems connected through standardized interfaces. Audio is acquired via I²S with DMAbased buffering and periodically stored on an SD card using burst-based writing to improve energy efficiency.Evaluation includes power measurements across sampling rates from 8 kHz to 192 kHz, field deployment with automated bioacoustic analysis, and a comparison with established PAM devices. Results show that storage operations dominate energy consumption, and that duty cycling significantly improves overall energy efficiency. The proposed sensing platform provides high-quality recordings suitable for automated acoustic inference and achieves competitive hardware performance while offering greater modularity, lower cost, and improved extensibility compared to many commercial PAM alternatives.The platform provides a foundation for future low-power edge AI processing and standardized wireless communication interfaces for distributed environmental monitoring.Publicationtype: Conference Paper24 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , TWIST: trains under weather, illumination, and seasonal timeRail transport is a cornerstone of sustainable and climate-friendly mobility. With a shift towards autonomous transport not only on the road but also on rails, AI-based track monitoring systems play an important role both for operating trains as well as for maintenance along the track. To build respective applications, we require high-quality datasets capturing real-world environmental variability not currently covered by any existing dataset containing images of trains. To address this gap, we create and publicly release TWIST, a domain-specific train image dataset collected across different seasons, comprising 38,000 real-world images at 640 × 480 resolution. TWIST captures trains under a wide range of weather and illumination conditions, including rain, snow, fog, low light, and night scenes, which are largely absent from existing public datasets and critically affect model robustness in deployment scenarios.To demonstrate the practical value of TWIST, we train a compact YOLOv8s object detection model from scratch and compare it against a pre-trained baseline. The model achieves a mean Average Precision (mAP@.50) of 60.4%, comparable to the baseline value of 60.8%, while substantially improving average precision for train detection (AP@.50) from 75.9% to 89.1%. These results highlight the importance of realistic data collection for robust railway monitoring.Publicationtype: Conference Paper12 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , At the edge of the heart: ULP FPGA-based CNN for on-device cardiac feature extraction in smart health sensors for astronauts(IEEE, 2026-06); ; ; ; The convergence of accelerating human spaceflight ambitions and critical terrestrial health monitoring demands is driving the unprecedented requirements for reliable and real-time feature extraction capabilities on extremely resource-constraint wearable health sensors. We present a ULP FPGA-based solution for real-time Seismocardiography (SCG) feature classification using Convolutional Neural Networks (CNNs). Our approach combines quantization-aware training with a systolic-array accelerator to enable efficient integer-only inference on the Lattice iCE40UP5K FPGA, which offers an ideal platform for battery-powered deployments - particularly in space environments - for their power efficiency and radiation resilience. The implementation achieves a validation accuracy of 98% while consuming only 8.55 mW of power, completing inference in 95.5 ms with minimal hardware resources (2,861 LUTs and 7 DSP blocks). These results demonstrate that fully on-device SCG-based cardiac feature extraction is feasible on resource-constrained hardware, enabling energy-efficient, autonomous health monitoring for astronauts in long-duration space missions.Publicationtype: Conference Paper24 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Short paper: simulation-based performance characterization of common NOR flash file systems(Association for Computing Machinery (ACM), 2026-05); ; NOR flash memory is widely used as storage in embedded systems as it offers dense non-volatile memory up to the gigabit range while being more reliable than NAND flash. To avoid the cost of a standalone flash controller in such small systems, specialized flash file systems are used to handle the limited write endurance and asymmetric program/erase characteristics. While many NOR file systems exist and are extensively deployed, there is a lack of systematic comparisons of their performance. In this paper, we present vflashsim, a simulator for capturing interactions between file systems and flash memory and for running representative benchmarks. Unlike evaluations on physical hardware, our simulation-based approach avoids prohibitively long erase latencies and cell wear. We implement a NOR flash model and, using vflashsim, evaluate four representative file systems, SPIFFS, LittleFS, YAFFS2, and NF2FS, in terms of write amplification, garbage-collection behavior, and RAM utilization. Our analysis shows how design choices and implementation bugs, such as excessive device scanning, unconditional sector erases, and a lack of garbage collection in allocation paths, cause substantial performance issues. Additionally, we identify how to provoke failure conditions in different file systems, which can result in unrecoverable crashes. Among the evaluated file systems, YAFFS demonstrates the best overall performance and reliability at the cost of high RAM usage, whereas a modified LittleFS remains the best option for more memory-constrained systems. Since performance trade-offs vary by workload and hardware constraints, no single file system is universally optimal. With vflashsim, we provide a tool that enables developers and users of flash file systems to analyze their behavior under diverse configurations and workloads.Publicationtype: Conference Paper20 13 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Undervolting experiments with LP-FPGAs in space(2026); ; ; Publicationtype: Conference Presentation6 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , GCStack+ and GCScaler+: Closing the calibration gap in GPU performance modelling for modern microarchitecturesGPU analytical performance models enable rapid design space exploration, but their accuracy depends heavily on the cycle-level simulators used for calibration. Existing tools such as GCStack and GCScaler are calibrated using Accel-Sim, whose execution pipeline is based on the NVIDIA Tesla architecture from 2006, featuring operand collector units, hardware scoreboards, and a Greedy-Then-Oldest warp scheduler. Recent reverse engineering studies have shown that modern NVIDIA GPUs eliminate operand collectors in favor of compiler-guided dependency management and adopt a Greedy-Then-Youngest warp scheduling policy. Although the simulator can still produce reasonable overall cycle counts, it often misattributes the underlying stalls in its CPI stacks and cache statistics. These inaccuracies are then passed onto any analytical model calibrated using them. In this paper, we revisit the calibration process used by analytical GPU performance models. We develop GCStack+, which generates CPI stacks and cache statistics using a simulator that reflects modern NVIDIA GPU microarchitecture, and GCScaler+, which uses these calibration inputs for performance prediction. The framework also adds support for Blackwell-generation instruction set, a protobuf-based trace format, and compiler-guided warp scheduling. Validated against an NVIDIA RTX 5070 (Blackwell) across nine Rodinia benchmarks, GCScaler+ reduces the geometric mean cycle prediction error from 18.2% (GCScaler) to 12.1%, at a 10.3× faster runtime than the simulator used during calibration.Publicationtype: Conference Paper31 17